Re: [weirds] Entity name searches

Yes, that is certainly doable but what is your initial query? "Give me
all your data", even if it is 1000 results at a time, is probably not
a good starting point.

-andy

On Mon, Nov 9, 2015 at 3:45 PM, Dave Piscitello
<dave.piscitello@icann.org> wrote:
> Can we clear up one question before we venture into extensions?
>
> Q: What is the purpose of such an extension versus the ability that is naturally available if one is able to collect structured data into a local data store?
>
> For my example, so long as I’m able to gather the 100,000 structured records, I and others will happily take structured data, incorporate these into SQL or other databases, and search to our hearts’ content. What has inhibited the kinds of _domain name_ investigations my example illustrates in the past was the combination of non-standard, unstructured data, rate limiting, and other completely non-RDAP related matters.
>
> Perhaps I’m missing something, but the extension(s) required to do what I suggested on a query level would require federated {authentication,authorization,auditing… } that would be quite remarkable objective for the numbers world and a monumental undertaking for the domain world, no?
>
>> On Nov 9, 2015, at 3:32 PM, Andrew Newton <andy@hxr.us> wrote:
>>
>> Ok. I understand the use case. It still requires an extension to RDAP.
>> That type of search never came up during the standardization process.
>>
>> At ARIN we do offer more targeted RESTful searches, but it has been my
>> experience they are used almost exclusively for data mining. And I
>> suspect that if left wide open, that's how it would be used in RDAP
>> for DNRs. So to get a "system" to work, there would need to be an
>> authorization system... and since we are talking multiple DNRs, that
>> would be federated authorization. And because there are multiple DNRs,
>> these queries would have to be run in parallel against each one (using
>> the bootstrap to find them). And then you have the problem that some
>> domain registries are thin. Those are a lot of problems to solve, and
>> I'm not sure who has the will to do it.
>>
>> -andy
>>
>> On Mon, Nov 9, 2015 at 12:44 PM, Dave Piscitello
>> <dave.piscitello@icann.org> wrote:
>>> More generally, if you are looking at 100,000 registration records that have
>>> been associated with spam, you’d want the ability to search or pivot on any
>>> data/string that establish relationships among the domain strings (e.g.,
>>> botnet DGA) but as importantly, _any_ element of the registration record
>>> that subsets share: POC, creation date, name server…
>>>
>>> On Nov 9, 2015, at 12:12 PM, Brian Mountford <mountford@google.com> wrote:
>>>
>>> Well, for instance, one might want to search the organization. When I do a
>>> WHOIS query for google.com, the contacts have an organization of Google Inc.
>>> I might want to search for all contacts with that organization.
>>>
>>> I can try to tailor the interpretation, but since the search string syntax
>>> does not allow for arbitrary suffix searching, it's not clear what tailoring
>>> is possible. Are you saying that I could take an entity name search string
>>> of "McB*" and interpret that as a search for names any of whose words begin
>>> with McB, so that it would find Joe McBride as well as McBurns Simpson? That
>>> seems to be playing pretty loose with the partial search string rules, since
>>> the RFC takes pains to define that syntax so precisely.
>>>
>>> Brian
>>>
>>> On Fri, Nov 6, 2015 at 4:47 PM, Andrew Newton <andy@hxr.us> wrote:
>>>>
>>>> Welcome to the world of internationalization, where the concept of a
>>>> last name vs a first name is not universal, and US ASCII is not
>>>> representative all the characters used.
>>>>
>>>> Since you know the data in your database best, you should taylor the
>>>> interpretation of the query input to that which works best with your
>>>> registry.
>>>>
>>>> On Sat, Nov 7, 2015 at 5:54 AM, Brian Mountford <mountford@google.com>
>>>> wrote:
>>>>>
>>>>> And only names? There is no provision for searching entities by address,
>>>>> etc.?
>>>>>
>>>>
>>>> I don't think that ever came up. It would require an RDAP extension. I
>>>> do question how useful such a thing would be. Why does anybody care
>>>> that a particular registry has contacts living on Mumford Lane in East
>>>> Westover? Are they searching all the registries for that information?
>>>> Do they need a telephone book instead?
>>>>
>>>> -andy
>>>
>>>
>>> _______________________________________________
>>> weirds mailing list
>>> weirds@ietf.org
>>> https://www.ietf.org/mailman/listinfo/weirds
>>>
>>>
>